It was possible to reach a state where a resampled burst would
overrun the transmit output buffer and corrupt the global
allocated signal vectors. The result was a segmentation fault
when attempting to access heap allocated signal vectors since
the pointers were garbage.
Whether the segfault occured or not appears to depend on the
memory location of the signal vector pointers, since it does
not occur on all systems.
Double buffer size to accomodate an incoming burst plus up to
a full chunk that may be remaining from the previous resampling
operation.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
With transceiver integration, the resampling filter files were
dropped. This created a working resampling implementation for
the USRP2 / N200, but with spectrum irregulaties that likely
caused issues at longer range operation. Simply reinsert the
filter files and modify the filter initialization to use them.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
The following option selects single daughterboard use on
side A only. Otherwise, the default setting is TX on A and
Rx on B.
./configure --with-singledb
Other options are possible, as per the following enum, but
for simplicity with configure options, there are only two
choices.
enum dboardConfigType {
TXA_RXB,
TXB_RXA,
TXA_RXA,
TXB_RXB
};
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
External reference selection was already compile-time
determined by a hard coded value. Make it selectable
as a configure option.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
With the unified transceiver, we need to select between
two device sample rates, GSM symbol rate and 400ksps for
resampler use. Make this selection available for configure
time options.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
Usage:
./configure --with-usrp1
Enable non-UHD USRP1 support through the gnuradio-based
driver. Requires gnuradio (libusrp) to be installed.
./configure --with-resample
Enable host-based 400ksps to 270.833ksps resampling.
Only supported for UHD devices. If not enabled, the GSM
sample rate is requested directly from the device.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
With the introduction of the B100, there is USB support
using UHD devices. The characteristics of the trasmit
side burst submissions are more reflective of the bus
type than the device or driver.
Use a fixed latency interval for network devices and the
adaptive underrun approach for USB devices - regardless
of driver or device type.
The GPMC based transport on the E100 appears unaffected
by either latency scheme, which defaults to network.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
The resampling transceiver is unified with the 52MHz
version. The option to resample 400ksps from the device
to a GSM appropriate 270.833ksps is enabled at compile
time with the following option.
./configure --with-resamp
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
The readSamples call does not return less than the
number of samples requested. Doing otherwise is a
fatal error. So on overruns, which are not fatal,
continue reading until the requested number of
samples is received.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
Move push and pull of buffers into a dedicated file. This will
allow us to swap out resampling, non-resampling, and possibly
floating point device interfaces while presenting a single
floating point abstration in the interface itself.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
Remove radio clock and vector interfaces into their own
files. This clears up and simplifies the radio interface
and, additionaly, prepares for a further split of the I/O
portion for optional resampler use.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
Pipe the following uhd message types to standard warning
levels (INFO, WARN, ERROR) respectively. Ignore fastpath
logging messages and, instead, catch them from the
asynchronous device interface.
enum type_t{
status = 's',
warning = 'w',
error = 'e',
fastpath= 'f'
};
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
Unused and causes the follwing errors on certain UHD versions.
"ValueError: unhandled clock configuration reference source: _external_"
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
If the master clock rate fails to set - this basically only happens
when the wrong transceiver is choosen for the particular device -
the error is fatal and the transceiver should exit. The clock rate
setting was previously never verified.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
This patch fixes some confusion in gain vs. attenuation
setting. The UHD device is controlled through gain
settings but OpenBTS represents gain in terms of
attenuation relative to maximum - 0 dB attenuation.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
UHD recently modified the E100 type name from 'usrp-e' to
'e100' causing the device make to fail. Remove device type
checking to keep things working with the older and newer
names.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
Commit e161523c (transceiver: simplify transmit power control)
changed transmit gain control to RF setting only. This was
appropriate for a WBX board with 25 dB of gain control, but
inappropriate for an RFX with fixed transmit gain.
RFX boards will regain the ability to set transmit
attenuation. Since gain is set on the RF side first,
reintroducing digital gain settings should have limited
overall effect on non-RFX daughterboards.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
* remote-cli: (22 commits)
Throw exception if usrp is not found. Don't crash transceiver.
Update OpenBTS init script
Correctly exit OpenBTS even if console is not started yet.
Add option to run OpenBTS in a failsafe loop.
Shutdown without core dump on a usual SocketError.
Remove Transceiver PID file on OpenBTS exit.
Save transceiver PID to a file and use it to kill old transceivers.
Correct comment for Server.WritePID.
Set FD_CLOEXEC flag on all sockets.
Stupid bug fixed - can't move stack variables initialization to a separate function.
Gracefully shutdown OpenBTS on transceiver timeout/error.
Make functions and variables in OpenBTS.cpp static to avoid their use outside of the file.
Remove PID file on program exit.
Implemented daemon mode, configured from config.
Better default name for OpenBTS control UNIX socket.
Initialize logger right after loading config file.
Use normal cout and cerr while we have only one thread.
Better comments for previous commit.
Catch SIGINT, SIGTERM and SIGHUP and shutdown the server gracefully.
Uncomment real BTS start/stop.
...
Receive buffer flush should continue to read until
either the desired number of packets has been read or
timeout, which means that the buffer has been emptied.
These are expected behaviours and should return true.
Ignore errors at this stage as the data and associated
metadata can be considered garbage and not worth
reporting. Actual error conditions will be caught
further downstream when useful data comes in.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
Transmit tuning was primarily setup for side A only. Some boards
- WBX - would still tune with improper channel arguments, though
receiver performance was disrupted.
Previous testing was primarily with single board on side A only
or dual configuration with side A transmit, so this bug largely
went undetected. Patch tested with RFX and WBX daughterboards
in single and dual configurations sides A and B.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
Bump abnormal asynchronous events - basically send errors -
up to ERROR level. These errors are dominated almost
entirely by underflow events, which should not be regularly
occuring.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
If no bursts were received over a long enough duration
then the threshold would roll into negative territory.
The energy detection is based on a comparison with the
squared threshold, so all handsets would become
effectively barred after a certain period of
inactivity.
In theory, this bug also exists in the mainline tree,
but there the daughterboard receive gain is fixed at
max, which always allows the ADC to generate sufficient
noise to trigger the energy dectector and keep the
system in a valid steady state.
To fix, simply add a negative value check like those
already in place for other locations.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
UHD will internally accept floats with a range of +/-1.0,
which corresponds to a 16-bit signed integer range of
apporximately +/- 32000. Set the default amplitude to .3,
which is a safe value agaist saturation elsewhere in the
transmit chain.
The non-UHD maximum amplitude is unchanged at 13500.
Remove digital gain control because it's unnecessary and
causes extra load on enbedded systems.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
This is primarily a minor refactor with the exception
of non-recoverable errors - notably if the receive times
out - which almost always requires a reload of the FPGA.
In these cases, quit without trying as resistance is
futile.
ERROR_TIMING: Soft restart of streaming
ERROR_UNHANDLED: Benign errors
ERROR_UNRECOVERABLE: Abandon ship
Non-recoverable behaviour has not been observed in recent
builds, but may exist in older (or future) configurations.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
Overnight testing shows that this shouldn't be required
in the majority of cases, but shit happens. Enabling
this forces transmit timing realignment at one minute
intervals. As a fallback method, timing slips not
caught by normal checks will be reset at the update.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
At startup, instead of flushing initial packets blindly,
send a stop streaming command, flush, and start. The same
procedure is used in the event of a runtime timestamp
validity error.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
The output of the modulator or resampler is scaled and
converted from floating point to fixed point. The scaling
factor is the leftover dB in RF attention (relative to max
transmit power), which is handled prior to the integer
conversion. This should work across all daughterboards and
non-UHD installations.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
* dboard:
Transceiver52M: add WBX, DBSRX, and single board support
Transceiver: add WBX, DBSRX, and single board support
Conflicts:
public-trunk/Transceiver/radioInterface.cpp
Previously this was referenced off the the ad9862
PGA with a range from 0 to -20 dB. Instead base
the attenuation factor on the maximum total RF
gain returned by the device.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
On a lapses of time monotonicity (and possibly other errors),
stop and restart the receive streaming with a buffer flush
in between. This is a cleaner replacement to the previous
clock reset with that didn't attempt to stop steaming.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
Type size_t was used in the UHD time_spec_t to integer
conversion, which would overflow at roughly 4 and a half
hours causing the sample buffer to error on timestamp
validity. Builds where size_t takes on 64-bits were not
affected by this bug.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
With UHD b4fc0d61bb6cbd1a5614745bab9aeb0abc22cb6f
Sample clock will reset to zero after an overrun. Earlier
versions may hang the FPGA, which is non-recoverable,
requiring a manual image reload or reboot.
If reset to zero, attempt to kick the sample clock to the
last properly received timestamp value. At this point,
there will be a timing continuity jump, which will drop
connections, but transmit and receive chains should be
aligned allowing for re-establishment.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
Push the ability to set thread priority out to the 52M
Transceiver interface, because that's where the thread
control exists.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
Use the same header files for the device and start moving
toward a commmon transceiver without so much redundant code.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
Shadow all gains and frequencies, which minimizes device access.
This allows the transceiver to variably control the device
settings.
Signed-off-by: Thomas Tsou <ttsou@vt.edu>
Harald Welte
Thomas Tsou